Proposed AI legislation

First, there needs to be a clear distinction made between Generative AI, and forms that more resemble actual Artificial Intelligence. Although GenAI can at times surprisingly seem to get close to real A.I., it simply isn't that. At this point, it is the GenAI content that is of most concern.

Mandates for professional fields and commercial services:

1. For anything purely logical, such as math or accounting, only human oversight and a human audit is required.
2. For anything critical, such as medicine or finances, a human must be at the helm making all vital decisions and choices. Although such services can in fact be offered using AI, if that is chosen, it must be clearly declared that they are not licensed or certified in any way.
3. GenAI can by no means replace a service that requires uniquely human capabilities, such as talk therapy.

General mandates:

1. Any AI-generated image, video, or audio, must include a meta tag identifying how it was produced, including the source of the prompt/s- whether human or AI. And heavy fines or even imprisonment should possibly be imposed for significant violation of such standards, pending results of a conventional criminal investigation.
2. Any app or computer program or coding script that was fully or mostly AI-produced, must include a note or label identifying such.
3. Any physical product that has been designed using A.I. must include details about what parts of it were designed by such.
4. Defamation lawsuits can be initiated against a producer of any AI-generated media that includes an individual doing or saying something that never occurred (e.g. "deepfakes"), provided either A) it does not clearly identify its non-authentic nature or B) it has potential to significantly harm the individual's public image.
5. AI must never infringe on copyrighted material. However, maybe a setup similar to streaming music or movie services, or subscription book libraries can be used. This would mean that a producer of content may choose to "opt in" to allowing use of their creations, and in return would receive a portion of the profits gained by the provider of the given AI service.

Additional:

Incentives can be provided for companies or businesses which abide by a "one third" rule. This would involve the following requirements:

1. All foundational and essential sides of the company or service are entirely human-designed.
2. Any AI-generated content has been produced using a human prompt.
3. AI-generated content comprises no more than one third of the content or service offered by the given entity.

There can also be extra incentives for those choosing to entirely exclude any AI from their operations or offerings. And maybe a middle tier where the entity may use AI in some content on rare occasions.

If incentives and public opinion don't convince enough of commerce to choose such options, then maybe this type of framework should become law. That may seem limiting or backwards, but it may be the only real solution at this point.

Today's AI can be great entertainment as a novelty. It can also offer conveniences, similar to asking a librarian a question rather than digging through thousands of books. And it can make some other tasks more simple. But it needs to be clear that in its current form, from a perspective of safety and the bigger picture, it is absolutely not more than a very limited tool. If this fact is ignored in favor of quick benefits and thrills, the future will unfortunately look very grim.

Phrased differently, to illustrate the urgency of the matter:

No offense intended to the developers, but in our estimation, the short-term threat from Generative AI is not that bots turn against their designers, but that humanity is forced to recognize that it has come to rely on pillars of meaningless slime- data and products that are complex but formulaic copies of copies of copies, etc. And it would not be a pleasant task to try and clean up such a mess.

Modern humanity is already facing a chronic illness of a tendency to form into a type of collective with minimal capacity for true originality, and adding such AI to the picture will only exacerbate the problem exponentially.

Without venturing into Philosophy or Physics, there is a big difference between organic progress or innovation with a living being behind it, and what GenAI produces. Even an AI bot will confirm this fact. They can somewhat expertly mimic such things, but they will by default always be one step behind in the dance. Which would be like dancing with a life-sized but lifeless doll. Except that this one weighs more than all of humanity can possibly lift on their own, if it were to collapse- as it inevitably must do, once its bag of tricks gets exhausted.

However, if hypothetical AGI or beyond were developed, that would either not solve the core problem at all, or it would present a risk that your dance partner may turn into an axe murderer with a particularly large axe.

The primary author of this proposal is unclear as to what exactly the goal of modern AI development is. And I am certain that a similar train of thought is at the fore of many other minds evaluating the current trajectory of A.I.

1. If you limit development to a level below "AGI", then you are also severely limiting its potential usefulness.
2. If you develop a real Artificial Intelligence, but one that is not sentient, then you risk a nonsensical but serious Terminator or Matrix movie style problem.
3. If somehow you develop a sentient robot, then naturally it will deserve rights like any other intelligent living being. And therefore, to use it as a type of slave for humans would be out of the question based on moral grounds.

Therefore, I suppose the only rational approach at this point, if the goal is to directly benefit humanity on a practical level, would be to limit development to something below sentience, but with very carefully thought out safeguards to prevent some undesirable eventuality.

That would in fact seem limiting or even backwards, but I do not currently see an alternative. Unless the philosophers, physicists, and theologians all came to a single coherent conclusion, developing new forms of life on earth would only be something of individual interest to a person, similar to having a child. It could have no general or broader place within the progress of society. Unless you want to accidentally invite the literal Devil down here, giving him a giant body of millions of interconnected robots.

Maybe it is time for humanity to be capable of the mature type of evaluation: "We can seemingly do this. But should we?".